Re: Database which allows object to be "child" of any other object

From: paul c <toledobythesea_at_oohay.ac>
Date: Thu, 09 Apr 2009 19:39:30 GMT
Message-ID: <S9sDl.22252$PH1.20288_at_edtnps82>


Bob Badour wrote:
> kuronekoyama_at_gmail.com wrote:
> ...
> If I am not mistaken, a name used for it before was "universal
> relation", but I could be wrong.
> ...

I'm not sure either because I seem to remember from years ago Ullman and maybe Kent writing about something more exotic than this quote from Date (pg. 420, Logic and Databases):

"Given a relation type RELATION {H}, where {H} is a heading, the universal relation of that type is the relation with heading {H} and body consisting of all possible tuples of type TUPLE{H}."

By "all possible tuples", I think he is excluding any additional constraints on a relation, beyond the ones that the attributes make, eg., if a candidate key has fewer attributes than {H} that constraint has to do with some relation value, not with the universal relation of the relationn type. Received on Thu Apr 09 2009 - 21:39:30 CEST

Original text of this message